Junqin Huang, L. Kong, Jiejian Wu, Yutong Liu, Yuchen Li, Zhe Wang
Mobile Internet enables a huge amount of access requests, leading to severe network congestion. To alleviate congestion in the transmission layer, lots of Congestion Control (CC) algorithms have been proposed recently in the research domain, which are specifically designed for various network environments. However, one of the teaching difficulties in mobile Internet education is to allow students to accurately choose the appropriate CC algorithm under the known or measurable network environment. In this paper, we propose a learning-based CC simulator for mobile Internet education, which provides intuitive suggestions to students on the CC algorithm selections via its learning ability in practical network environments. Our simulator consists of three key modules: the network data module, learning module, and CC module. It has built-in several default CC algorithms and supports students' customized algorithms. The performance of the proposed simulator is evaluated on the implemented simulator prototype with both real and simulated network links. Evaluation results show that the simulator can dynamically select proper CC algorithms in the light of network environments to achieve higher throughput, which benefits students in understanding the working mechanisms of CC algorithms intuitively.
{"title":"Learning-based congestion control simulator for mobile internet education","authors":"Junqin Huang, L. Kong, Jiejian Wu, Yutong Liu, Yuchen Li, Zhe Wang","doi":"10.1145/3477091.3482760","DOIUrl":"https://doi.org/10.1145/3477091.3482760","url":null,"abstract":"Mobile Internet enables a huge amount of access requests, leading to severe network congestion. To alleviate congestion in the transmission layer, lots of Congestion Control (CC) algorithms have been proposed recently in the research domain, which are specifically designed for various network environments. However, one of the teaching difficulties in mobile Internet education is to allow students to accurately choose the appropriate CC algorithm under the known or measurable network environment. In this paper, we propose a learning-based CC simulator for mobile Internet education, which provides intuitive suggestions to students on the CC algorithm selections via its learning ability in practical network environments. Our simulator consists of three key modules: the network data module, learning module, and CC module. It has built-in several default CC algorithms and supports students' customized algorithms. The performance of the proposed simulator is evaluated on the implemented simulator prototype with both real and simulated network links. Evaluation results show that the simulator can dynamically select proper CC algorithms in the light of network environments to achieve higher throughput, which benefits students in understanding the working mechanisms of CC algorithms intuitively.","PeriodicalId":305393,"journal":{"name":"Proceedings of the 16th ACM Workshop on Mobility in the Evolving Internet Architecture","volume":"162 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120861133","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The growing popularity of connected vehicles is driving the development of a new class of latency-sensitive applications. The applications can benefit from edge clouds but they need timely responses to their requests. For applications based on video and other sensor data, this necessitates high burst throughput from the network, which results in a challenging scheduling problem since applications have diverse request sizes, rates, and deadlines. While the problem of managing latency sensitive bursty traffic has been considered in data centers, wireless networks use variable bit rates, which further exacerbates the problem. In this paper, we first use simple examples to show that traditional network fairness and quality-of-service (QoS) solutions are not appropriate for latency sensitive applications using cloud services at the wireless edge. We then sketch a possible solution that addresses the challenges of combining Service Level Agreements that focus on burst transfers with a network scheduler that is resource-aware. Finally, we use a simplified network, namely Wi-Fi based emulation, to show how our approach enables both the network administrator and applications to meet their goals.
{"title":"Meeting connected vehicle application requirements: it's not just about bandwidth","authors":"Kwame-Lante Wright, P. Steenkiste","doi":"10.1145/3477091.3482762","DOIUrl":"https://doi.org/10.1145/3477091.3482762","url":null,"abstract":"The growing popularity of connected vehicles is driving the development of a new class of latency-sensitive applications. The applications can benefit from edge clouds but they need timely responses to their requests. For applications based on video and other sensor data, this necessitates high burst throughput from the network, which results in a challenging scheduling problem since applications have diverse request sizes, rates, and deadlines. While the problem of managing latency sensitive bursty traffic has been considered in data centers, wireless networks use variable bit rates, which further exacerbates the problem. In this paper, we first use simple examples to show that traditional network fairness and quality-of-service (QoS) solutions are not appropriate for latency sensitive applications using cloud services at the wireless edge. We then sketch a possible solution that addresses the challenges of combining Service Level Agreements that focus on burst transfers with a network scheduler that is resource-aware. Finally, we use a simplified network, namely Wi-Fi based emulation, to show how our approach enables both the network administrator and applications to meet their goals.","PeriodicalId":305393,"journal":{"name":"Proceedings of the 16th ACM Workshop on Mobility in the Evolving Internet Architecture","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121825603","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A. Banchs, M. Fiore, Andres Garcia-Saavedra, M. Gramaglia
The success of the upcoming 6G systems will largely depend on the quality of the Network Intelligence (NI) that will fully automate network management. Artificial Intelligence (AI) models are commonly regarded as the cornerstone for NI design, as they have proven extremely successful at solving hard problems that require inferring complex relationships from entangled, massive (network traffic) data. However, the common approach of plugging ‘vanilla’ AI models into controllers and orchestrators does not fulfil the potential of the technology. Instead, AI models should be tailored to the specific network level and respond to the specific needs of network functions, eventually coordinated by an end-to-end NI-native architecture for 6G. In this paper, we discuss these challenges and provide results for a candidate NI-driven functionality that is properly integrated into the proposed architecture: network capacity forecasting.
{"title":"Network intelligence in 6G: challenges and opportunities","authors":"A. Banchs, M. Fiore, Andres Garcia-Saavedra, M. Gramaglia","doi":"10.1145/3477091.3482761","DOIUrl":"https://doi.org/10.1145/3477091.3482761","url":null,"abstract":"The success of the upcoming 6G systems will largely depend on the quality of the Network Intelligence (NI) that will fully automate network management. Artificial Intelligence (AI) models are commonly regarded as the cornerstone for NI design, as they have proven extremely successful at solving hard problems that require inferring complex relationships from entangled, massive (network traffic) data. However, the common approach of plugging ‘vanilla’ AI models into controllers and orchestrators does not fulfil the potential of the technology. Instead, AI models should be tailored to the specific network level and respond to the specific needs of network functions, eventually coordinated by an end-to-end NI-native architecture for 6G. In this paper, we discuss these challenges and provide results for a candidate NI-driven functionality that is properly integrated into the proposed architecture: network capacity forecasting.","PeriodicalId":305393,"journal":{"name":"Proceedings of the 16th ACM Workshop on Mobility in the Evolving Internet Architecture","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126196374","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A. Abubakar, Nishanth R. Sastry, Mohamed M. Kassem
For disconnected communities, the associated cost of backhaul network infrastructure is seen as one of the most significant challenges of accessing connectivity services. To help mitigate this challenge, we propose the Internet Island Architecture. The Architecture provides room for a large class of so-called ``sharing economy'' applications such as temporary labour, dating service, market services and transport services supported by local connectivity alone. We show the feasibility of the system and discuss the advantages of the proposed architecture
{"title":"Internet islands: first class networked communities in isolated regions","authors":"A. Abubakar, Nishanth R. Sastry, Mohamed M. Kassem","doi":"10.1145/3477091.3482763","DOIUrl":"https://doi.org/10.1145/3477091.3482763","url":null,"abstract":"For disconnected communities, the associated cost of backhaul network infrastructure is seen as one of the most significant challenges of accessing connectivity services. To help mitigate this challenge, we propose the Internet Island Architecture. The Architecture provides room for a large class of so-called ``sharing economy'' applications such as temporary labour, dating service, market services and transport services supported by local connectivity alone. We show the feasibility of the system and discuss the advantages of the proposed architecture","PeriodicalId":305393,"journal":{"name":"Proceedings of the 16th ACM Workshop on Mobility in the Evolving Internet Architecture","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132142246","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：